Top 5 Problems of Managed Video Surveillance / SaaS
Today in 2009, Managed video is becoming one of the most overhyped trends in the video surveillance industry.
A number of factors drive interest. The concept of reducing on-site setup and eliminating deployment of DVR/NVRs is quite attractive. Further fueling interest is its potential to generate recurring revenue streams rather than installation revenue. In the background is the general IT trend of migrating software applications to on-line services.
In talking with industry professionals, I see a lot of noise and unclarity about how managed video works and how competitive it is (and will be) versus traditional video surveillance solutions.
While I believe in the long term potential (last year I selected managed video as one of 3 top emerging technology trends), it's not a fully mature technology. As such, there are many problems that will constrain its use for years to come.
Here are the top five:
- Last mile bandwidth limitations
- Cost/complexity of deploying/integrating cameras/recorders on-site
- Advantages are not as compelling as they are in other applications
- Generally weak business case compared to traditional solutions
- Vendor promotion is driving the hype
- Plus - Free offerings will limit growth
- Recording continues to be done locally (on-site). This is often referred to as managed video because the video is still in the same place as it traditionally was but a third party provider manages access and administration of this video.
- Recording is moved off-site. This is often referred to as hosted video. The video is no longer stored at the customer's location. Rather, it is streamed over an IP network to the third party provider who stores the video.
The terms used and the approaches to managed video vary. Let's start by contrasting it to the current mainstream approach.
Over 99% of video surveillance deployments consist of hardware, software and storage deployed and managed at the customer's site. If a user wants to monitor or administer the system they need to access the internal network of the customer.
By contrast, the service model transfers some or all of these elements to a third party provider that is physically separate from the customer. There are 2 general approaches to this:
Last mile bandwidth limitations
Central to the future of Video Surveillance as a Service is eliminating on-site recording. To do this, sufficient bandwidth must be available to transmit video from on-site cameras to off-site recording. Once done, systems save the cost and maintenance of an on-site server/recorder.
Proponents, such as Axis, frequently claim that bandwidth limitations are being eliminated. They cite Moore's Law and the abundance of Wide Area Network bandwidth.
The problem is in last-mile bandwidth, the network connection serving almost all offices and buildings. While massive amounts of bandwidth is available connecting cities, very little bandwidth is available servicing individual offices and buildings. Most sites are lucky to have a few megabits of bandwidth available connecting to the Internet.
Making matters worse is that off-site storage of surveillance video depends on upstream bandwidth availability. Almost all streaming is going from the cameras on-site upstream to the 3rd party provider off site. Most last mile networking technologies have lower upstream bandwidth than downstream bandwidth (cable modem, ADSL, cellular systems, etc.). For instance, you may have 5 Mb/s download but only 1Mb/s upload. The 5 Mb/s download is irrelevant. Only the 1 Mb/s upload is useful for transmitting surveillance video to the off-site provider.
Because of this, providers offering off-site storage target 1 - 4 cameras per site with very low frames rates (usually 1-3 fps). Contrast this with on-site recording where 16 cameras and 15 fps per camera is ordinary. As such, these off-site services are at a significant disadvantage on 2 key features.
Last mile bandwidth will increase but likely slow and only over a period of years. Proponents like to talk about fiber to the home (FTTH) and 4G networks. However, after years of hype, FTTH is still very narrowly deployed. Moreover, 4G networks are not designed to handle continuous streaming of low value surveillance video. Wireless carriers would almost certainly charge significant overage fees for subscribers that attempted to continuous stream surveillance video. This would make such use uneconomic for all but mobile and applications without wireline alternatives.
Cost/complexity of deploying/integrating cameras/recorders on-site
With video surveillance as a service, devices still need to be placed physically at the customer's site (unlike other SaaS applications like e-mail or CRM). The issue remains whether it is placing cameras only on site or cameras plus an on-site recorder.
The problem arises because the use of dynamic private IP addresses are common within internal networks (LANs). Remote video surveillance services cannot directly connect to cameras using such addresses. Either an on-site installer must change configurations to allow remote access or special software needs to be pre-loaded onto the cameras to automatically allow remote access.
While some require installers to make changes on-site (such as ByRemote), others pre-load software that automatically enable remote access (such as Axis STS). The cost and complexity of doing on-site changes makes it almost inevitable that for these services to scale, special software will need to be pre-loaded.
All software enabling automatic remote connection is proprietary and none of them today work for multiple manufacturers. As such, this limits use to a single manufacturer's products. This is beneficial to the manufacturer but limiting to the market. For instance, Axis STS technology helps promoting Axis products but at Axis' higher prices it weakens the business case compared to traditional analog + DVR solutions.
Advantages are not as compelling as they are in other applications
In IT, two of the most powerful drivers of SaaS are applications that (1) require access to the application outside of the internal LAN and (2) involve significant tuning and optimization.
For instance, sales people are primary users of CRM systems. They absolutely need to access these systems on the road as they are making calls. Enabling such access with traditional systems requires internal IT people to allow remote access into the Internal network. This is costly, increases some security risks and is still limiting for the employee (as they often need to use VPNs that are slow and require set-up). By contrast, SaaS CRM (like Salesforce) works just like logging into any other website.
In video surveillance, most use is from inside of the company's network (whether at the local site or from a remote side part of the same network). While there are benefits for some users, it is not as critical as it is for other enterprise applications where system users are routinely on the road.
The other major general SaaS advantage that is not as important in video surveillance is the cost of tuning and optimization. An e-mail server may require complex setup and ongoing maintenance to back up, expand system, etc. This is much different than a DVR where the application is pre-loaded, the database is integrated and the recordings are automatically deleted in a circular fashion. Vendors routinely generate business cases assuming massive ongoing support with skilled IT technicians. This is simply not the case with DVRs.
Generally weak business case compared to traditional solutions
At current price points and functionalities offered, most of the managed video offerings are not very financially attractive. From discussions with a half dozen providers, the average charge per month per camera is about $30 USD for off-site recording at 1-3 fps 4CIF video for 30 days (cameras and installation of cameras are charged separately).
By contrast, traditional deployments cost about $15 per month. An 8 channel DVR costs approximately $4,000 USD. Even assuming a short 3 year lifecycle, that's $13.80 USD per camera per month. 8 channel DVRs over the first 3 years rarely need much maintenance. If a maintenance call costs $250 USD, that's about $1 per camera per month.
Proponents will push the value of the safety of off-site storage, RAID and easier remote access. The reality, though, is that very few people are willing to double their cost to obtain these features.
I have no doubt that prices will decrease for these services but until they do, the service will remain a niche.
Vendor promotion is driving the hype
Like all over-hyped trends, this is being driven by vendors and amplified by trade magazines. While this factor is common with all over-hyped trends, it is important to appreciate the specific drivers here.
Vendors want managed video to succeed because it holds the promise of improving their financial position. Investors more highly value companies with recurring rather than installation revenue. Plus, given the state of the economy, vendors hope that monthly fees will be more attractive than up front costs. This, of course, does not mean that it will succeed or when it will succeed.
Secondly, and specifically by Axis, managed video is being promoted as a means to eliminate the cost gap of IP against analog video for small camera count deployments. Of course, to accomplish this requires the elimination of on-site recorders. However, given the pricing of these services and the limitation of the offerings, it's not very economically attractive even for more than a few cameras per site.
Free Offerings Will Limit Growth
In this super-small camera count segment, free offerings will emerge to limit growth of subscription services. A number of Western companies are OEMing such a solution from StarVedia while DLink has recently announced a similar offering (plus see our note on DLink in a recent weekly review).
These offerings will primarily provide simple remote viewing (the automatic remote connection software described above) but will do it at no monthly cost. For users with small sites and simple systems, this should be good enough for many especially at no recurring cost.
Conclusion
While I believe manged video use will grow, it will be slowly over time. Keep these issues in-mind as you evaluate using managed video.